Audio Description (AD), also known as Video Description in North America, is a service that provides a separate narration for the blind or partially sighted. Used in media, performing arts and museums it provides a description of the action on the stage or screen during natural gaps in the dialogue. Audio description can be delivered in many ways depending on the environment it is used in and the technologies chosen. For TV & DVD the audio is supplied either as a separate audio track that mixes the description audio with the main audio track or in the case of some TV services as a separate narration only track that is mixed in the consumers TV giving them control over the balance of the mix. In cinemas and museums the user will have a control unit and headset that allows only them to hear the narration track provided. For all live performances where the timing can vary considerably then the narration script can either be spoken by a voice artist live or each segment of audio triggered manually. The recent improvements in the quality of computer generated voices has led to new opportunities. Now it is possible to use a range of voices to match accent, gender and age to the performance. This gives greater flexibility in the authoring process and can save time and increase productivity.

There are many applications where audio recordings have to be transcribed to a written document or where it would be a very useful and searchable resource, if it could be done economically.  Business and government meetings, lectures, presentations and all forms of media and broadcast content. There have been many attempts to use automatic speech recognition (ASR) or Speech to Text systems to do this but often the resulting output needs too much manual correction to be cost effective against 100% manual transcription except in some niche markets with very limited vocabulary sets. The advance of neural-network or AI based ASR systems in the last few years has started to change this and several systems are now available as web services. By integrating audio handling, audio filtering, content specific lexicons and easy to use editing tools the process can become more cost effective and 100% web deployed.

Source files for subtitling and closed captioning  come in a huge range of types and standards. The delivery formats for Subtitles and Captions within the media or broadcast signal are even more complex and vary between regions of the globe. Input files may be in  STL,PAC,SRT.SUB,890,RTF,TXT,CAP,CC,SCC, EBU-TT, SMPTE-TT, other Timed Text formats. These then need to be encoded into a delivery format that may be DVB, Teletext, or Closed caption in CEA608 or EIA708.
Once encoded the subtitle or caption data must to be added to a media file or broadcast signal in, VBI, VANC, MXF, MPEG-2, DVB TS, SMPTE436M, OP-47, WebVTT, EBU-TT-D or other emerging web formats.

Once Subtitles and Closed Captions have been encoded and embedded into one of a wide range of  media files they are almost impossible to access. If a video file containing Subtitles or Captions has to be edited for re-use this valuable information is almost always lost and is recreated from scratch for the new edit. Subtitles and Captions can always be extracted from a media file or even a broadcast signal in a wide range of formats, VBI, VANC, MXF, MPEG-2, DVB TS, SMPTE436M, OP-47 and turned back into standard subtitle file formats. This reduces the cost of re-use and recreates a valuable asset which is also a useful source of searchable metadata as a timed word list.